Prediction of apricot export volume using artificial intelligence

This study aims to determine the most appropriate predicting method for the export volume of apricot products, which is one of the essential export items in Turkey. While deciding the predicting strategy, the most appropriate prediction method is found as a result of comparing the predictions with Artificial Neural Networks (ANN), Seasonal Autoregressive Integrated Moving Average (SARIMA), and Extreme Gradient Boosting (XGBoost) methods. In this study, 2002-2020 General Trade System (GTS) Foreign Trade monthly data is obtained from the Turkish Statistical Institute (TURKSTAT). In which seasonality is observed, Apricot data is estimated using AR, ARIMA, SARIMA, ANN, and XGBoost methods for 2020. As a result, XGBoost achieves better prediction accuracy than SARIMA and ANN, according to the impact of the performance measure.